How do frictions in markets in developing countries affect the spatial distribution of eco- nomic activity? Does agricultural income continue to play an important role in determining the economic opportunities of urban firms, or do global capital and goods markets make local income irrelevant? We provide new evidence on these classic questions in development economics. We find that positive shocks to rural agricultural income induce employment growth in urban areas, with growth concentrated in the manufacturing sector. In testing between demand and capital channels that could explain such manufacturing growth, our evidence suggests that agricultural surplus increases growth in manufacturing firms by providing capital, suggesting both that such firms are capital constrained and that local informal capital markets can successfully intermediate between local rural surplus and urban investment opportunities.
Given these findings, it is necessary for economists to move beyond their focus on the relationship between agriculture and the rural non-farm economy. It should be of little
Table 1.10:Effect of rainfall on manufacturing growth by market access
Low Trans Costs High Trans Costs Pooled
Rain (sum) 0.044 0.025 0.046 (0.013)*** (0.023) (0.012)*** Rain * High TC -0.026 (0.016) High TC -0.215 (0.128) N 245 277 522 r2 0.47 0.38 0.39
∗p<0.10,∗∗p<0.05,∗∗∗p<0.01. Dependent variable is urban district employment growth in manufacturing establishments, in logs. All regressions with robust standard errors and state-year fixed effects, clustered at the
state level. Rain is defined as -1×the absolute value of standard deviations of rainfall minus 1.25, considered the optimal amount of rainfall, summed over the five year period before the end of the Economic Census period. Period 0 thus corresponds to the five year window 1993-1997 and period 1 to 2000-2004. High TC is a variable representing above median transportation costs from the district to the nearest city of 500,000 people (2001). Control variables for all regressions are period, land, urban and rural population, and share of employment in banking in 1990.
surprise to development economists that the performance of urban economies is linked to that of nearby rural areas. Migration, consumption and capital flows are but three examples of the linkages between urban and rural economies.
These findings have implications for many debates in development policy. First, policies that increase agricultural production may also contribute to industrialization, although we must be cautious when applying findings from shocks to long term changes in productivity. Second, other policies that increase rural income, such as the Mahatma Gandhi National Rural Employment Guarantee Scheme (NREGS), may also increase industrialization. It should be noted that the increase in manufacturing employment that we observe occurs entirely in urban areas. This suggests that although manufacturing firms in rural areas may be credit constrained, returns in urban areas appear sufficiently high to overcome the proximity that rural firms have to the source of the income shock. Differences between returns to rural and urban manufacturing are certainly a topic for further inquiry.
between agricultural earnings and investment opportunities in manufacturing. Further research is needed to understand the channels by which agricultural surplus can fund urban manufacturing growth and the reasons that capital is able move from rural to urban areas within a district but is sufficiently localized to produce local growth, rather than being invested nationally. Although informal markets appear to successfully facilitate manufacturing growth, it is necessary to understand the efficiency with which they are able to fund entrepreneurship in order to judge how well they can substitute for formal financial institutions.
Chapter 2
Natural Resource Wealth and Local
Politics
1
2.1
Introduction
At least as far back as Thomas Malthus, economists have been concerned with the role of natural resources in economic development. Sachs and Warner (1995) launched a modern literature into the issue, demonstrating that natural resource wealth was correlated across countries with underdevelopment. That natural resource dependence often coexists with poor governance is widely known; however, we are still far from understanding the channels by which mineral wealth could undermine institutional quality, and the conditions under which this occurs. As pointed out by Isham et al. (2005), while Chad and the Democratic Republic of the Congo may suffer from clear curses of natural resources, the United States and Norway provide equally strong examples of countries with long and consistent growth records whose institutions were not undermined by vast natural resource wealth.
Natural resources do not appear at random, and they tend to be fixed over time, making inference about their effects difficult. Places rich in natural resources may lack other natural advantages; for example, they may be mountainous or remote. Counterfactual stories for places rich in natural resources are inherently challenging to defend. Rather than attempt
the very difficult task of constructing a counterfactual political environment for a resource rich location, this paper compares natural resource rich locations with each other over time. We use price shocks to provide plausibly exogenous variation to the value of subsurface natural resources in a given location. The context is India, a nation which as a whole is not highly dependent on point-source natural resources2, but has many regions which produce a wide range of natural resources, including coal, iron, gold and various other minerals.
This provides an ideal context for examining the impact of mineral wealth on political outcomes. First, by using subnational and subregional variation, we are able to compare outcomes in regions that share basic political institutions. Second, by interacting global prices changes with the presence of mineral deposits, we generate plausibly exogenous variation in the value of subsoil mineral resources. India contains considerable internal variation in mineral deposits but overall production is small relative to world totals (Indian Bureau of Mines, 2011), eliminating the possibility of reverse causality from local political outcomes to global mineral prices. Third, Indian law defines the national and state governments to be claimants of all mineral taxes and royalties. We are therefore able to rule out one of the primary mechanisms by which natural resources are thought to undermine political outcomes: significant increases in local government revenue.
The starting point for this paper is the observation that changes in natural resource wealth influence political outcomes through many, often highly tangled and overlapping, channels. We propose and test three mechanisms by which mineral wealth may influence political outcomes. First, the selection effect posits that greater mineral wealth will change the composition of the candidate pool in local elections. Second, the election effect suggests that conditional on the candidate pool, mineral wealth changes the likelihood of electoral victory of certain types of candidates. Third, in the moral hazard effect politicians change their behavior in office in response to changes in the value of subsoil resources. We find strongest evidence for the election effect: winning candidates are much more likely than other candidates to have criminal cases against them. Our incumbency results also suggest
2We exclude agriculture from this study.
that the probability of victory for certain types of candidates is increasing in mineral wealth. At this point we cannot identify the underlying mechanism for these results. As will be discussed later in this paper, potential candidate and party responses to mineral wealth prevent us from attributing such results to changes in voter behavior alone.
This paper provides some of the first microeconomic causal evidence of a direct effect of natural resource wealth on local political outcomes. When a region experiences a boom in the value of subsurface natural resources, politician criminality increases significantly. In particular, we find that winning candidates are significantly more likely to be implicated in criminal cases. Margins of victory increase and incumbents are more entrenched: the vote share of both local incumbents and the state ruling coalition is increasing in the global prices of local mineral deposits. Thanks to the exceptional nature of Indian laws regarding mineral royalties, we are able to provide what we believe to be the first evidence of political consequences of natural resource wealth in the absence of significant changes to local revenue.
Despite competitive elections, criminality is common among politicians in democracies around the world. This is particularly true in developing countries, with potentially large negative consequences for welfare (Chemin, 2012). Postulated reasons for such a prevalence of criminals range from the advantage of criminals in intimidating voters (Aidt et al., 2011) to a lack of political competition (Banerjee et al., 2012a). Our findings, while not conclusive, point away from a lack of choice: winners are more likely to be criminal than the rest of the field of candidates. In subsequent sections we discuss some of the reasons why criminals may have electoral advantages in the presence of high mineral wealth.3
The rest of the paper is organized as follows. Section 2.2 reviews the existing literature examining the linkages between natural resources and political outcomes. Section 2.3 gives background information on mining in India and Indian political organization. Section 2.4 describes the sources of data and the construction of variables. Section 2.5 presents the
3Such electoral advantages may include greater appeal to voters, but not necessarily. For example, criminal politicians may exert greater effort to win elections if they are then better able to capture rents once in office.
conceptual framework that guides our empirical investigation. Section 2.6 explains our empirical strategy. Section 2.7 presents and discusses the results. Section 2.8 concludes.